Mean Reward on ALFWorld
0.767Mean RewardNO-COMP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| NO-COMPBackbone=qwen3.5-flash, Inference Protocol=Inference-free2026.05 | 0.767 | — | |
| AGORABackbone=qwen3.5-flash, Inference Protocol=Inference-free2026.05 | 0.767 | 100 | |
| TruncBackbone=qwen3.5-flash, Inference Protocol=Inference-free2026.05 | 0.767 | 100 | |
| NO-COMPBackbone=gpt-5-mini, Inference Protocol=Inference-free2026.05 | 0.767 | — | |
| AGORABackbone=gpt-5-mini, Inference Protocol=Inference-free2026.05 | 0.633 | 83 | |
| TruncBackbone=gpt-5-mini, Inference Protocol=Inference-free2026.05 | 0.633 | 83 | |
| Floor-K2Backbone=qwen3.5-flash, Inference Protocol=Inference-free2026.05 | 0.6 | 78 | |
| Floor-K2Backbone=gpt-5-mini, Inference Protocol=Inference-free2026.05 | 0.5 | 65 | |
| LongLinguaBackbone=qwen3.5-flash, Inference Protocol=Inference-free2026.05 | 0.467 | 100 | |
| LongLinguaBackbone=gpt-5-mini, Inference Protocol=Inference-free2026.05 | 0.467 | 93 | |
| HiAgentBackbone=qwen3.5-flash, Inference Protocol=LLM-call per step2026.05 | 0.4 | 86 | |
| ACONBackbone=qwen3.5-flash, Inference Protocol=LLM-call per step2026.05 | 0.4 | 86 | |
| HiAgentBackbone=gpt-5-mini, Inference Protocol=LLM-call per step2026.05 | 0.4 | 80 | |
| AgentDietBackbone=qwen3.5-flash, Inference Protocol=LLM-call per step2026.05 | 0.333 | 71 | |
| AGORABackbone=gpt-4o-mini, Inference Protocol=Inference-free2026.05 | 0.333 | 111 | |
| ObsMaskBackbone=qwen3.5-flash, Inference Protocol=Inference-free2026.05 | 0.3 | 64 | |
| NO-COMPBackbone=gpt-4o-mini, Inference Protocol=Inference-free2026.05 | 0.3 | — | |
| TruncBackbone=gpt-4o-mini, Inference Protocol=Inference-free2026.05 | 0.267 | 89 | |
| ObsMaskBackbone=gpt-5-mini, Inference Protocol=Inference-free2026.05 | 0.267 | 53 | |
| AgentDietBackbone=gpt-5-mini, Inference Protocol=LLM-call per step2026.05 | 0.267 | 53 | |
| Floor-K2Backbone=gpt-4o-mini, Inference Protocol=Inference-free2026.05 | 0.233 | 78 | |
| ACONBackbone=gpt-5-mini, Inference Protocol=LLM-call per step2026.05 | 0.233 | 47 | |
| AgentDietBackbone=gpt-4o-mini, Inference Protocol=LLM-call per step2026.05 | 0.167 | 71 | |
| ACONBackbone=gpt-4o-mini, Inference Protocol=LLM-call per step2026.05 | 0.167 | 71 | |
| ObsMaskBackbone=gpt-4o-mini, Inference Protocol=Inference-free2026.05 | 0.1 | 43 | |
| LongLinguaBackbone=gpt-4o-mini, Inference Protocol=Inference-free2026.05 | 0.1 | 43 | |
| HiAgentBackbone=gpt-4o-mini, Inference Protocol=LLM-call per step2026.05 | 0.1 | 43 | |
| SelCtxBackbone=qwen3.5-flash, Inference Protocol=Inference-free2026.05 | 0.033 | 7 | |
| SelCtxBackbone=gpt-4o-mini, Inference Protocol=Inference-free2026.05 | 0 | 0 | |
| SelCtxBackbone=gpt-5-mini, Inference Protocol=Inference-free2026.05 | 0 | 0 |